From ce4a32fe52ef09d8f5ff1dd22c001110902b60a2 Mon Sep 17 00:00:00 2001 From: toma Date: Wed, 25 Nov 2009 17:56:58 +0000 Subject: Copy the KDE 3.5 branch to branches/trinity for new KDE 3.5 features. BUG:215923 git-svn-id: svn://anonsvn.kde.org/home/kde/branches/trinity/kdelibs@1054174 283d02a7-25f6-0310-bc7c-ecb5cbfe19da --- khtml/SAFARI_MERGE | 161 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 161 insertions(+) create mode 100644 khtml/SAFARI_MERGE (limited to 'khtml/SAFARI_MERGE') diff --git a/khtml/SAFARI_MERGE b/khtml/SAFARI_MERGE new file mode 100644 index 000000000..7ef67e134 --- /dev/null +++ b/khtml/SAFARI_MERGE @@ -0,0 +1,161 @@ + +This is a short document with the progress of patches in the Safari codebase +being merged into khtml. + +its ordered by directory and by files. Add here the parts that cannot be +immediately merged and need further negotiation. + +Its recommended to add a //FIXME_SAFARI comment to places which need merging +in the khtml sourcecode. + +=========================================================================== +# css +=========================================================================== + +- the font changes (except for the konq-body property) are still not merged. + They are partly not easily portable and I don't understand some parts of them. + a small part is merged but still unused (separate font sizes for fixed fonts) + +- The Apple specific parts with their DPI settings. + +- most of the html4.css changes are not in. I have already added the + @konq-quirks part + +=========================================================================== +# dom +=========================================================================== + +- dom_doc.cpp / .h: preferred Stylesheet stuff. how does this correlate + to the changes in css_stylesheet.h / cpp ? <- It's independent (lars). + requires merging of the impl in xml/html first + +- css_stylesheet.cpp / .h addition of preferredStyleSheet et all, but + appears to be unused + +- weird change in dom_doc.cpp: doctype() + +- dom_string.h: removal of virtual destructor can't be merged, BIC. + doesn't make sense anyway, as one might want to inherit from it. + if it is a speed problem more stuff has to be handled with DOMStringImpl* + +- html_document.cpp / .h setPolicyBaseURL for cookie handling. euhm? + +=========================================================================== +# ecma +=========================================================================== + +TODO: +- ScriptInterpreter domObjectsPerDocument + getDOMDocumentNode + +To investigate: +- kjs_navigator (PluginBase refcount ; ProductSub) + hmm, why is e.g. the ref/deref stuff in APPLE_CHANGES? +- scope changes (kjs_events.cpp, DOMNode::pushEventHandlerScope) +- Image object: DOM::Document doc -> QGuardedPtr doc; + (why? Or rather: should this also be done in all other domobject-wrappers?) +- window.scroll == scrollto or scrollby ?? +- ScheduledAction: ObjectImp -> Object. Sounds good - why did we have an imp? refcount problem? + +Needs other stuff before it can be applied: +- Identifier change (needs kjs2) (includes lengthPropertyName change) + +=========================================================================== +# html +=========================================================================== + +- dtd.cpp: changes regareding ID_TEXT, ID_FORM, A and FONT left out + +- html_formimpl.cpp/.h: many changes skipped. will need to integrate the + "activate first submit" button part. + +- html_headimpl.cpp: changes for FOUC in external style loading + +- html_image/html_form: name / id lookup cache skipped. want to do that + for all elements. + +- html_miscimpl.cpp: case insensitive name / id matching + +- html_parser: reapplying of dropped attributes. reported to be obsolete + +=========================================================================== +# misc +=========================================================================== + +- decoder.cpp/.h: changes in APPLE_CHANGES #ifdef that make decoder + being a null wrapper just using unicode directly instead of working + on charset specific 8bit data. + +- htmltags.in: removal of <-konqblock>, not sure if that should ever + be merged. + +=========================================================================== +# rendering +=========================================================================== + +done so far: + +- RenderArena support. Usage of the arena is disabled for now. Needs + testing before I'll enable it. + +- renamed print methods to paint + +- RendeLayer code is used. Needs to be updated to apples latest version. + +- Handling of block within inlines via continuations is done. + +changes that need to be merged: + +- bidi.h/cpp: lot's of changes to white space handling and others. + probably needs the merge of the tokenizer. + +- font.cpp/h: some apple specific changes. Addition of a genericFamily enum, + IMO to speed up style stuff. Should IMO be separated out into a font_mac.cpp + and font_x11.cpp + +- repaint has gotten a bool immediate + +- render_applet.cpp: small #ifdef APPLE changes + +- render_box.cpp: + paintBackground has some fixes for background position in + table cells and repeat handling. Need to examine in detail + + calcWidth has some code for min and max width constraints + calcHeight has a hack for "height: 100%". Not sure I understand it. + +- render_container: + some hack to avoid an infinite recursion on one of the evil tests. + Since I fixed anonymous table insertion, I doubt it is still needed. Needs testing. + +.... + + +=========================================================================== +# xml +=========================================================================== + +- dom2_eventsimpl.cpp: computeLayerPos, requires mering of rendering/ + +- dom2_rangeimpl.cpp: crateContextualFragmet.. merging needed + +- dom_docimpl.cpp: + renderArena support.. merging needed. + lowercase attribute names.. requires confirmation + preferred style sheet / delay recalcing style selector till + all external stylesheets are loaded. + +- dom_nodeimpl.cpp: rendererNeedsClose. I don't fully understand this + seems to be only required for the avoid-unstyled-elements-flicker + restoreState -> QStringList change. will merge. + setLayouted(false) calls in attach(). this should imho be handled + in the rendering tree. needs feedback. + +- dom_nodeimpl.cpp / .h: setMalformed() hack. + isInline() fixes. + previous/nextrenderer addition + +- dom_stringimpl.cpp: toLengthArray alternative implementation not merged + +- dom_textimpl.cpp: setNodeValue: whitespace nuking needs merged!! + +- xml_tokenizer.cpp/.h: completely skipped. unfinished work. -- cgit v1.2.1